mirror of
https://github.com/rsyslog/rsyslog.git
synced 2025-12-19 11:30:43 +01:00
template procesing/json: performance optimization
This patch improves the performance of important parts of template JSON processing.
This commit is contained in:
Rainer Gerhards
parent
d9870419a9
commit
6cf38c0ce7
156
runtime/msg.c
156
runtime/msg.c
@ -7,7 +7,7 @@
|
||||
* of the "old" message code without any modifications. However, it
|
||||
* helps to have things at the right place one we go to the meat of it.
|
||||
*
|
||||
* Copyright 2007-2020 Rainer Gerhards and Adiscon GmbH.
|
||||
* Copyright 2007-2022 Rainer Gerhards and Adiscon GmbH.
|
||||
*
|
||||
* This file is part of the rsyslog runtime library.
|
||||
*
|
||||
@ -3290,13 +3290,17 @@ finalize_it:
|
||||
}
|
||||
|
||||
|
||||
/* Encode a JSON value and add it to provided string. Note that
|
||||
* the string object may be NULL. In this case, it is created
|
||||
* if and only if escaping is needed. if escapeAll is false, previously
|
||||
* escaped strings are left as is
|
||||
/* Helper for jsonAddVal(), to be called onces we know there are actually
|
||||
* json escapes inside the string. If so, this function takes over.
|
||||
* Splitting the functions permits us to make some performance optimizations.
|
||||
* For further details, see jsonAddVal().
|
||||
*/
|
||||
static rsRetVal
|
||||
jsonAddVal(uchar *pSrc, unsigned buflen, es_str_t **dst, int escapeAll)
|
||||
static rsRetVal ATTR_NONNULL(1, 4)
|
||||
jsonAddVal_escaped(uchar *const pSrc,
|
||||
const unsigned buflen,
|
||||
const unsigned len_none_escaped_head,
|
||||
es_str_t **dst,
|
||||
const int escapeAll)
|
||||
{
|
||||
unsigned char c;
|
||||
es_size_t i;
|
||||
@ -3304,39 +3308,69 @@ jsonAddVal(uchar *pSrc, unsigned buflen, es_str_t **dst, int escapeAll)
|
||||
unsigned ni;
|
||||
unsigned char nc;
|
||||
int j;
|
||||
uchar wrkbuf[100000];
|
||||
size_t dst_realloc_size;
|
||||
size_t dst_size;
|
||||
uchar *dst_base;
|
||||
uchar *dst_w;
|
||||
uchar *newbuf;
|
||||
DEFiRet;
|
||||
|
||||
for(i = 0 ; i < buflen ; ++i) {
|
||||
assert(len_none_escaped_head <= buflen);
|
||||
/* first copy over unescaped head string */
|
||||
if(len_none_escaped_head+10 > sizeof(wrkbuf)) {
|
||||
dst_size = 2 * len_none_escaped_head;
|
||||
CHKmalloc(dst_base = malloc(dst_size));
|
||||
} else {
|
||||
dst_size = sizeof(wrkbuf);
|
||||
dst_base = wrkbuf;
|
||||
}
|
||||
dst_realloc_size = dst_size - 10; /* some buffer for escaping */
|
||||
dst_w = dst_base;
|
||||
memcpy(dst_w, pSrc, len_none_escaped_head);
|
||||
dst_w += len_none_escaped_head;
|
||||
|
||||
/* now do the escaping */
|
||||
for(i = len_none_escaped_head ; i < buflen ; ++i) {
|
||||
const size_t dst_offset = dst_w - dst_base;
|
||||
if(dst_offset >= dst_realloc_size) {
|
||||
const size_t new_size = 2 * dst_size;
|
||||
if(dst_base == wrkbuf) {
|
||||
CHKmalloc(newbuf = malloc(new_size));
|
||||
memcpy(newbuf, dst_base, dst_offset);
|
||||
} else {
|
||||
CHKmalloc(newbuf = realloc(dst_base, new_size));
|
||||
}
|
||||
dst_size = new_size;
|
||||
dst_realloc_size = new_size - 10; /* some buffer for escaping */
|
||||
dst_base = newbuf;
|
||||
dst_w = dst_base + dst_offset;
|
||||
}
|
||||
c = pSrc[i];
|
||||
if( (c >= 0x23 && c <= 0x2e)
|
||||
|| (c >= 0x30 && c <= 0x5b)
|
||||
if( (c >= 0x30 && c <= 0x5b)
|
||||
|| (c >= 0x23 && c <= 0x2e)
|
||||
|| (c >= 0x5d /* && c <= 0x10FFFF*/)
|
||||
|| c == 0x20 || c == 0x21) {
|
||||
/* no need to escape */
|
||||
if(*dst != NULL)
|
||||
es_addChar(dst, c);
|
||||
*dst_w++ = c;
|
||||
} else {
|
||||
if(*dst == NULL) {
|
||||
if(i == 0) {
|
||||
/* we hope we have only few escapes... */
|
||||
*dst = es_newStr(buflen+10);
|
||||
} else {
|
||||
*dst = es_newStrFromBuf((char*)pSrc, i);
|
||||
}
|
||||
if(*dst == NULL) {
|
||||
ABORT_FINALIZE(RS_RET_OUT_OF_MEMORY);
|
||||
}
|
||||
}
|
||||
/* we must escape, try RFC4627-defined special sequences first */
|
||||
switch(c) {
|
||||
case '\0':
|
||||
es_addBuf(dst, "\\u0000", 6);
|
||||
*dst_w++ = '\\';
|
||||
*dst_w++ = 'u';
|
||||
*dst_w++ = '0';
|
||||
*dst_w++ = '0';
|
||||
*dst_w++ = '0';
|
||||
*dst_w++ = '0';
|
||||
break;
|
||||
case '\"':
|
||||
es_addBuf(dst, "\\\"", 2);
|
||||
*dst_w++ = '\\';
|
||||
*dst_w++ = '"';
|
||||
break;
|
||||
case '/':
|
||||
es_addBuf(dst, "\\/", 2);
|
||||
*dst_w++ = '\\';
|
||||
*dst_w++ = '/';
|
||||
break;
|
||||
case '\\':
|
||||
if (escapeAll == RSFALSE) {
|
||||
@ -3347,31 +3381,35 @@ jsonAddVal(uchar *pSrc, unsigned buflen, es_str_t **dst, int escapeAll)
|
||||
/* Attempt to not double encode */
|
||||
if ( nc == '"' || nc == '/' || nc == '\\' || nc == 'b' || nc == 'f'
|
||||
|| nc == 'n' || nc == 'r' || nc == 't' || nc == 'u') {
|
||||
|
||||
es_addChar(dst, c);
|
||||
es_addChar(dst, nc);
|
||||
*dst_w++ = c;
|
||||
*dst_w++ = nc;
|
||||
i = ni;
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
es_addBuf(dst, "\\\\", 2);
|
||||
*dst_w++ = '\\';
|
||||
*dst_w++ = '\\';
|
||||
break;
|
||||
case '\010':
|
||||
es_addBuf(dst, "\\b", 2);
|
||||
*dst_w++ = '\\';
|
||||
*dst_w++ = 'b';
|
||||
break;
|
||||
case '\014':
|
||||
es_addBuf(dst, "\\f", 2);
|
||||
*dst_w++ = '\\';
|
||||
*dst_w++ = 'f';
|
||||
break;
|
||||
case '\n':
|
||||
es_addBuf(dst, "\\n", 2);
|
||||
*dst_w++ = '\\';
|
||||
*dst_w++ = 'n';
|
||||
break;
|
||||
case '\r':
|
||||
es_addBuf(dst, "\\r", 2);
|
||||
*dst_w++ = '\\';
|
||||
*dst_w++ = 'r';
|
||||
break;
|
||||
case '\t':
|
||||
es_addBuf(dst, "\\t", 2);
|
||||
*dst_w++ = '\\';
|
||||
*dst_w++ = 't';
|
||||
break;
|
||||
default:
|
||||
/* TODO : proper Unicode encoding (see header comment) */
|
||||
@ -3379,12 +3417,54 @@ jsonAddVal(uchar *pSrc, unsigned buflen, es_str_t **dst, int escapeAll)
|
||||
numbuf[3-j] = hexdigit[c % 16];
|
||||
c = c / 16;
|
||||
}
|
||||
es_addBuf(dst, "\\u", 2);
|
||||
es_addBuf(dst, numbuf, 4);
|
||||
*dst_w++ = '\\';
|
||||
*dst_w++ = 'u';
|
||||
*dst_w++ = numbuf[0];
|
||||
*dst_w++ = numbuf[1];
|
||||
*dst_w++ = numbuf[2];
|
||||
*dst_w++ = numbuf[3];
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
if(*dst == NULL) {
|
||||
*dst = es_newStrFromBuf((char *) dst_base, dst_w - dst_base);
|
||||
} else {
|
||||
es_addBuf(dst, (const char *) dst_base, dst_w - dst_base);
|
||||
}
|
||||
finalize_it:
|
||||
if(dst_base != wrkbuf) {
|
||||
free(dst_base);
|
||||
}
|
||||
RETiRet;
|
||||
}
|
||||
|
||||
|
||||
/* Encode a JSON value and add it to provided string. Note that
|
||||
* the string object may be NULL. In this case, it is created
|
||||
* if and only if escaping is needed. if escapeAll is false, previously
|
||||
* escaped strings are left as is
|
||||
*/
|
||||
static rsRetVal ATTR_NONNULL(1, 3)
|
||||
jsonAddVal(uchar *const pSrc, const unsigned buflen, es_str_t **dst, const int escapeAll)
|
||||
{
|
||||
es_size_t i;
|
||||
DEFiRet;
|
||||
|
||||
for(i = 0 ; i < buflen ; ++i) {
|
||||
const uchar c = pSrc[i];
|
||||
if(! ( (c >= 0x30 && c <= 0x5b)
|
||||
|| (c >= 0x23 && c <= 0x2e)
|
||||
|| (c >= 0x5d /* && c <= 0x10FFFF*/)
|
||||
|| c == 0x20 || c == 0x21)
|
||||
) {
|
||||
iRet = jsonAddVal_escaped(pSrc, buflen, i, dst, escapeAll);
|
||||
FINALIZE;
|
||||
}
|
||||
}
|
||||
if(*dst != NULL) {
|
||||
es_addBuf(dst, (const char *) pSrc, buflen);
|
||||
}
|
||||
finalize_it:
|
||||
RETiRet;
|
||||
}
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user